Robust linear MIMO receivers: a minimum error-rate approach

This paper looks at the linear reception of spatially multiplexed signals across MIMO channels. We address the problem of robustness in the presence of detrimental effects such as correlation and Ricean components. We consider the error-rate performance of MIMO linear filters, as these can be used in purely linear receivers, or as part of each stage in successive interference canceling (SIC) receivers. We know from multiuser detection theory that minimum error-rate (MER) linear receivers significantly outperform minimum mean-square error (MMSE) receivers when correlation is high, however no direct method exist to design the MER receiver simply. We derive a scheme allowing a closed-form approximate solution to this problem. The solution is a good approximation to the true MER receiver upon fulfillment of a certain, easily checkable, channel-related condition. The algorithms are derived first for the two-input many-output case. A generalized scheme is provided for the case of arbitrary number of inputs and outputs. The performance gain compared to that MMSE is evaluated for various correlated and Ricean channels and transmit power allocation strategies.